Disequilibrium Decomposition and Breakdown of Muscovite in High P- T Gneisses, Betic Alpine Belt (Southern Spain)

American Mineralogist, Volume 78, pages 158-177, 1993

Disequilibrium decomposition and breakdown of muscovite in high P- T gneisses, Betic alpine belt (southern Spain)

ANTONIO GARCiA-CASCO, ANTONIO SANCHEZ-NAVAS, RAFAEL LUIS TORRES-RoLDAN Departamento de Mineralogia y Petrologia and LA.G.M., Consejo Superior de Investigaciones Cientificas, Universidad de Granada, Fuentenueva sin, 18002-Granada, Spain

ABSTRACT Muscovite from anatectic leucogneisses of the Torrox gneiss complex (Betic alpine belt, southern Spain) has an unusually wide range in composition. Primary matrix crystals are Si-rich (up to 6.66 atoms pfu, normalized to 20 0 atoms and 4 OH) and contain abundant structurally controlled biotite + quartz intergrowths that range in size from a few nano- meters to tens of micrometers. Diffusion halos, depleted in Si, Fe, and Mg, developed in muscovite around these intergrowths. Also, a continuous compositional trend of decreasing Si, Fe, Mg, and Ti and increasing AI, Na, and K was formed by primary and recrystallized matrix muscovite grains, the latter forming the end product of this evolution, with Si contents as low as 6.14 atoms pfu. Inspection of TEM images of the finer biotite intergrowths reveals that (1) these intergrowths represent a variety of frozen stages of growth, with interfaces that changed from coherent and semicoherent to incoherent upon coarsening, and (2) that biotite-quartz nucleation was probably influenced by deformation defects in the host muscovite. A third textural variety of muscovite appears as large, homogeneous, pegmatitic grains with low Si contents (6.13 atoms pfu), which underwent partial low-P breakdown to fibrolite + andalusite + potassium feldspar + biotite. The primary and recrystallized muscovite crystals do not bear indications of breakdown. The textures and compositional features associated with the decomposition of the phengitic components of primary muscovite, and with the breakdown of the pegmatitic variety, suggest that the overstepping of the respective reaction boundaries was rapid. This is consistent with independent estimates of the P-T-t trajectory of the Torrox gneiss complex, which was characterized by rapid and marked decompression of approximately 10 kbar at high T (from> 10 to 2 kbar, and from 650 to 600 °C) and was followed by very rapid cooling (>200 °C/m.y.) at low P. Phengitic decomposition took place during decompression and deformation, resulting in a wide range of intermediate compositions, the intergrown biotite + quartz assemblage, and the diffusion halos. The reaction textures and the residual composition of primary muscovite are interpreted in terms of a change in the rate-limiting step from surface- to diffusion-controlled growth upon coarsening of product phases. The absence of potassium feldspar in the product phase assemblage, which conflicts with the stable equilibrium relationships and mass balance calculations, indicates that K diffused to the matrix. Although decomposition proceeded probably far from equilibrium (i.e., strongly overstepped), it is suggested that it followed the theoretical behavior expected for muscovite solid solution, indicating that P substantially influences the Si content of high-grade muscovites similarly to what is well known in lower T environments. The metastable persistence of muscovite outside the subsolidus stability of muscovite + quartz is explained by a combination of (1) a P-T path that evolved near the muscovite-out isograd, (2) high rates of cooling and decompression at low P, and (3) moderate displacements in the P- T location of the equilibrium reaction due to compositional differences of the phases involved, most probably the fluid phase.

INTRODUCTION cribed to two main solid solutions, the paragonite-muscovite series, involving the NaK_1 exchange component, Because of their widespread occurrence in many meta- and the phengite series, involving the Tschermak (tk, morphic and igneous rocks, muscovite solid solutions are SiMgAI_2) exchange component, which relates muscovite important in the description and interpretation of natural to the leucophyllite component K2Mg2AI2Sig02o(OH)4. systems. The major deviations of muscovite from its ide- These substitutions are known to be strongly influenced al composition, K2AI4AI2Si602o(OH)4, are normally as- by intensive and extensive parameters and to display an 0003-004X/93/0 102-0 158$02.00 158 GARCIA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES 159 antipathetic behavior in muscovite solid solutions (see cases ranging from 6.1 to 6.3 atoms pfu (either meta- Guidotti, 1984, for a review). Natural phengitic musco- morphic, e.g., Evans and Guidotti, 1966; Guidotti, 1973, vite also includes Fe2+ and Fe3+, involving the FeMg_1 1978a; Tracy, 1~78; Fletcher and Greenwood, 1979; Hol- and the Fe3+AI_l exchanges (Wise and Eugster, 1964), the daway et aI., 1988; or igneous, e.g., Guidotti, 1978b; Mil- latter less well constrained and documented. ler et aI., 1981; Lee et aI., 1981; Kistler et aI., 1981; Price, The deviation of muscovite from ideal composition as 1983; Monier and Robert, 1986b; Sevigny et aI., 1989), a result of the Tschermak substitution has been studied (2) the positive dPldT slopes of reaction isopleths, in in natural, experimental, and theoretical model systems. combination with P- T paths dominated by cooling, or (3) For a given system, composition, and phase assemblage, favorable kinetics under slow changes in P-T conditions a consistent decrease in the Si content of natural mus- and availability of fluids. However, that heterogeneous covite occurs with decreasing P and increasing T (e.g., muscovite populations often develop within single sam- Ernst, 1963; Cipriani et aI., 1971; Guidotti and Sassi, ples, and even within individual crystals, in cooling plu- 1976; Guidotti, 1984). The system composition and phase tonic bodies (e.g., Miller et aI., 1981; Speer, 1984; Monier assemblage also exert an important control on the extent et aI., 1984; Konings et aI., 1988) indicates that musco- of substitution: at constant P- T conditions, muscovite vite equilibration is a sluggish process under high- to me- from strongly aluminous compositions, bearing one or dium-temperature conditions. more AI-saturating phases, is more AI-rich and Si-poor The subsolidus upper stability of muscovite in Si02- than muscovite from less aluminous compositions that saturated rocks is described by the univariant equilibri- coexist with potassium feldspar (Guidotti, 1973; Miya- um (KASH system) shiro and Shido, 1985). The change in muscovite composition due to the Tschermak exchange within the limiting equilibrium. assemblage Bt + Qtz + Kfs + H20 muscovite quartz (mineral abbreviations after Kretz, 1983) has been investigated by Velde (1965, 1967), Monier and Robert (1986a), aluminum silicate potassium feldspar and Massonne and Schreyer (1987). The model net-transfer reaction (J.B. Thompson, 1982b) describing the di- which is commonly considered as the low-T limit (iso- variant equilibrium (KMASH system) is grad) for high-grade rocks. In spite of the discontinuous nature of Reaction 2a, muscovite can be found coexisting 3K2AI6Si602o(OH)4 + 6SiMgAl_2 with Kfs + Qtz + aluminum silicate across a zone of muscovite Tschermak's substitution persistence in high-T rocks subjected to prograde metamorphism at low to intermediate P. This zone might re- potassium feldspar quartz sult from (1) kinetic factors (Kerrick et aI., 1991), (2) the heterogeneous nature of fluids (e.g., Speer, 1982) as Re-

action 2a is displaced to lower T at the condition P H20 < biotite Ptot(e.g., Kerrick, 1972), or (3) variations in the compo- written in terms of the Tschermak exchange vector to sition of the solid phases. Experimental and theoretical emphasize the increasing Al contents of the micas on de- studies on Reaction 2a show that, under low to moderate composition (J.B. Thompson, 1979; A.B. Thompson, overstepping of the reaction boundary, muscovite should 1982). Si isopleths for Reaction 1 (further referred to as rapidly react to completion (e.g., Ridley and Thompson, phengitic muscovite decomposition) have low positive 1986; Schramke et aI., 1987; Kerrick et aI., 1991) and dPldT slopes (Velde, 1965, 1967; Massonne and Schrey- that the displacements of Reaction 2a in the P-T space er, 1987), indicating a strong pressure effect on the sub- due to variations in the composition of muscovite would stitution, which is in good agreement with natural meta- not account for this zone of persistence. The reaction morphic muscovite. The presence of reaction textures resulting from Re- action 1 helps constrain P-T paths (Massonne and muscovite Tschermak's substitution quartz Schreyer, 1987). Given the geometry of Si isopleths, reaction textures and associated compositional heterogeneities in muscovite are often found in high-P blueschist aluminum silicate potassium feldspar and eclogite facies metamorphic rocks that were subjected to decompression (e.g., Heinrich, 1982; Saliot and biotite Velde, 1982; Franz et aI., 1986; Evans and Patrick, 1987). (2b) Indications of instability of high- T phengitic muscovite are not so common; however, Ferrow et ai. (1990) have which includes the phengite components in muscovite presented evidence for exsolution of a 10-A phyllo- and the production of a ferromagnesian phase (i.e., bio- silicate, perhaps celadonite, within granitic muscovite. The tite, KMASH system), occurs at a point only a few de- lack of reaction textures caused by Reaction 1 in high -T grees higher than Reaction 2a (A.B. Thompson, 1982), muscovite is the result of (1) its low Si content, in most but this displacement is counterbalanced by the effect of 160 GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-TGNEISSES

tonic units (Fig. 1). The TGC crops out (-- 3 km2 and CSJ 1lalaguides > 500 m thick) at the bottom of the upper Alpujarride B Alpujarrides unit in the area (Torrox unit; Fig. 1), beneath a thick []] sequence of amphibolite facies graphitic schists. Both the Nevado-Filabrides graphitic schists and the TGC gneisses experienced a similar deformational and metamorphic history during the Alpine orogeny. The latest penetrative synmetamorphic deformation event resulted in the dominant megascopic mylonitic foliation (D2, of Cuevas et aI., 1989). Assem- blages in the system A-F-M-Ti in the graphitic metapelites adjacent to the gneissic body include pre-S2 Ms + St + Grt + Bt + Ky + Rt, syn-S2 fibrous sillimanite + St + Bt + 11m, and post-S2 And + Bt + 11m :t Crd. All D Plio-Quaternary three aluminum silicate polymorphs coexist in most sam- ~ Vinuela Formation ples, and a complex mineral growth and dissolution his- D Flysch Unit tory can be deduced from the textures and compositions of the minerals. Fibrous sillimanite + Bt + 11m and And + Bt + 11m pseudomorphs after staurolite, Bt + Ms :t D Velez-Malaga PI after gamet, Crd + Bt after garnet and staurolite, and m Salares Torrox replacement of kyanite by andalusite indicate that lower ~ Gneiss Complex P assemblages evolved from higher P assemblages by . Competa means of staurolite and garnet breakdown reactions upon Sierra Tejeda I ~ decompression. Comparable examples of frozen disequi- Nerja librium are common in other Alpujarride sequences ~ Mediterranean Sea (Loomis, 1976, 1979; Torres-Roldan, 1974, 1981). Uni- form cooling ages for numerous minerals in several iso- Fig. 1. Geologic sketch map of the Velez Malaga-Sierra Te- topic systems (cooling ages cluster around 21 :t 2 Ma; jeda massif in the central Betic zone of the Betic Cordilleras. Location of TGC (Torrox gneiss complex) is shown. Lithologic Zeck et aI., 1989, 1992; Monie et aI., 1991a, 1991 b) sug- sequences are not differentiated within units. Based on an un- gest that rapid cooling followed rapid decompression, published map by R.L. Torres-Roldan (see Torres-Roldan, 1979; precluding complete reequilibration at low pressure. Elorza and Garcia-Duenas, 1981; Sanz de Galdeano, 1989, for Within the TGC the main type of rock is a medium- an explanation of these regional terms and further geologic in- grained granitic leucogneiss. In addition to quartz, sodic formation). plagioclase, and potassium feldspar, the other major phases are muscovite and biotite (up to 10-20%) and, in Na in the micas and feldspars (e.g., Chatterjee and Froese, some cases, garnet « 5%). Accessory minerals include 1975). Indications for Reaction 2 (either 2a or 2b, further abundant apatite and tourmaline, whereas zircon, ilmen- referred to as muscovite breakdown) are not common in ite, rutile, kyanite, sillimanite, andalusite, and dumorti- high-P, medium-to-high T rocks subjected to decom- erite are less commonplace. Three main varieties of leu- pression, indicating that the P-Tpaths followed by high-P cocratic gneiss can be differentiated in terms of their rocks are normally dominated by cooling at intermediate textural, phase-assemblage, and outcrop characteristics: to low pressure. muscovite biotite garnet banded gneiss, augen and por- In this paper we document what appears to be a rare phyritic muscovite biotite gneiss, and muscovite garnet natural case ofphengitic muscovite decomposition (prod- biotite aplites containing muscovite, tourmaline, and uct assemblage Bt + Qtz) and breakdown (product as- spessartine garnet with pegmatitic sizes. The porphyritic semblage And or fibrous sillimanite + Kfs + Bt) in high muscovite biotite gneiss contains centimeter-sized alu- P- T anatectic leucogneisses (Torrox gneiss complex, minous enclaves (restites) with a primary assemblage of southern Spain). Complex reaction microtextures and wide Bt + Ky + Rt + Grt + Ap + graphite, which is over- compositional heterogeneities developed within these printed by sillimanite, andalusite, ilmenite, and musco- muscovites as the result of a tectonic setting that caused vite. Nongranitic rocks also appear interbedded with the the original high P- T assemblages to undergo rapid near- gneisses, including graphitic pelitic layers with variable isothermal decompression at high temperature. amounts of muscovite, quartz, biotite, garnet, aluminium silicates, feldspars, ilmenite, rutile, graphite, apatite, zircon, and tourmaline. All these rock types occur as strong- FIELD AND PETROLOGICAL SETTING ly foliated layers or lensoid bodies, with the exception of The Torrox gneiss complex (TGC) is a heterogeneous some porphyritic muscovite biotite gneisses and the ap- gneissic body that is located in the central segment of the lites. The latter form concordant layers and centimeter- Betic Cordilleras, about 50 km east of Malaga; the area to meter-sized pods and dikes with relationships that are includes several Alpujarride, Malaguide, and flysch tec- diffuse to sharply crosscutting, suggesting an origin by GARCIA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES 161 short-range melt segregation. Due to strong deformation parallel to the foliation of the rock, are interpreted as and alpine metamorphism (which includes small to mod- pegmatitic segregations that formed close to the H20- erate amounts of partial melting), as well as its lithologic saturated granitic solidus. These pegmatitic muscovite heterogeneity, the origin of the TGC as a whole is still samples bear partial pseudomorphs consisting of coarse- uncertain. Alternative hypotheses range from a metased- grained And or fibrous sillimanite + Kfs + Bt, which are imentary derivation to a pre alpine anatectic gneiss. not observed in the primary and recrystallized matrix muscovite.

SAMPLE DESCRIPTION The muscovite populations from four banded musco- ANALYTICAL PROCEDURES vi te biotite garnet gneiss sam pIes were studied. Phase as- An automated Cameca SX50 electron microprobe semblages in the four samples are fairly homogeneous. In (University of Granada), operated at 20 kV, was used to addition to the main granitic matrix assemblage Qtz + determine mineral compositions. Data were reduced us- PI + Kfs, these samples consist of Ms + Bt + Grt and ing a 4>-p-zprocedure supplied by the manufacturer (see abundant, coarse, rounded, apatite crystals (up to 0.5 mm Pouchou and Pichoir, 1985, for details), using simple ox- in size). Potassium feldspar also forms megacrysts that ides (AI203, Fe203, MnTi04, MgO) and silicates (albite, include idiom orphic plagioclase crystals with inverse orthoclase, wollastonite) as calibration standards. De- zoning and xenomorphic cores. These megacrysts show tailed wavelength scans indicated no appreciable amounts intense deformation and sigmoid tails. Matrix biotite is of other elements such as Ba, Sr, and CI in the micas. F not abundant (ca. 5%), and occurs parallel to the main might be present in low amounts, but could not be reli- foliation defined by muscovite. Medium- to fine-grained ably estimated on a routine basis. In the case of musco- garnet is always present « 5%), has corroded rims, and vite, special care was taken to locate the electron beam is partly replaced by medium- to fine-grained aggregates on areas where no intergrown phases appeared at the scale of Ms + Bt :t PI :t Qtz and by late retrograde (green) of optical and backscattered electron (BSE) images. For biotite. Accessory phases include tourmaline and very muscovite, biotite, and feldspars, operating parameters scarce fine-grained rutile and ilmenite. No aluminum sil- were optimized to avoid alkali loss, after the results of a icate polymorphs were detected in the matrix of the an- number of experiments, as illustrated in Figure 2. In ac- alyzed samples. cordance with the results of these experiments, both Na In all samples muscovite is more abundant than bio- and K were always counted first (simultaneously) and tite. With the exception of scarce retrograde aggregates beam current and probe sizes were combined so as to use after potassium feldspar, three main textural types of current densities of ca. 0.8 nA ~m-2, for which no appre- muscovite habits can be distinguished: primary deformed ciable decrease in counting statistics was expected during matrix crystals, recrystallized matrix grains, and large their counting times (15-25 s). However, large current pegmatitic crystals. Primary muscovite crystals have tex- densities (up to 20 nA ~m-2) had to be used for small tural characteristics that suggest equilibrium with the en- biotite intergrowths (5 nA, 0.25-~m2 probe size). Repre- closing granitic gneiss. In the samples investigated, de- sentative mica compositions are listed in Table 1 (see formation textures and microfabrics are commonplace, Appendix Tables 1 and 21 for the complete set of 66 spot but muscovite still has a medium- to coarse-grained analyses of muscovite studied and representative com- booklet habit (0.5-2 mm), and it has clean, sharp con- positions of coexisting phases, respectively). tacts with feldspars. Associated with primary crystals are The TEM images were obtained with a Zeiss EM10C biotite-quartz (+ minor occasional rutile) intergrowths (University of Granada), operated at 100 keY. An objec- on the optical to TEM scale (10-30 ~m to 10-3 ~m, see tive aperture of 40 ~m, corresponding to a minimum dhkl below), oriented parallel to (001) planes of the host mus- value of 0.35 nm, was selected to allow adding 001 dif- covite. Recrystallized muscovite grains have resulted from fracted and 000 transmitted beams to achieve a compro- shear deformation and grain-size reduction of primary mise between amplitude contrast and phase contrast in crystals. They appear as randomly oriented fine-grained the images. TEM specimens, selected from specially pre- grains «0.5 mm) associated with primary deformed pared thin sections, were thinned using Ar ion-beam crystals. Some of these recrystalliz€d grains also contain techniques to obtain electron-transparent edges. Quali- thin intergrowths of biotite, similar to but much less tative analyses were performed with a Jeol 1200-EX abundant than those within primary grains. Pegmatitic equipped with a LINK energy-dispersive X-ray spec- crystals, up to 5 cm long and 1.5 cm across, appear less trometer (University of Cadiz) operated at 120 ke V and commonly within the muscovite biotite garnet banded with a spot size of 100 nm to help with the identification gneisses, and particularly in one of the samples selected ofTEM-scale intergrowths within primary muscovite. for detailed study (T506). These larger muscovite crystals are typically associated with coarse tourmaline and form 1 A copy of Appendix Tables 1 and 2 may be ordered as Doc- ument AM-93-514 from the Business Office, Mineralogical So- the core of sparse augen-like structures, with a micro- ciety of America, 1130 Seventeenth Street NW, Suite 330, granular outer rim of quartz, potassium feldspar, and al- Washington, DC 20036, U.S.A. Please remit $5.00 in advance bite. The augen structures, which are deformed and lie for the microfiche. --,~-

GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES

2.0

Biotite T316 b , Crystal 1.5 o 'G'I U) c '+' 5 E:: 1.0 +bl) ~+ QJ ~0.5

4 0.0 4.0 4.5 5.0 5.5 6.0

r.IAl +!8JAl

Fig. 3. (a) Multicationic diagram including relevant mica end- members to illustrate the continuous nature of the coupled cation substitutions underlying the compositional variations within the investigated muscovite populations. Abbreviations are explained in the text and Figure 12, except eas, which stands for the eastonite end-member [K2[6JAI2Mg4AI4Si402o(OH)4].(b) En- larged section of the plot in (a), including a regression on the primary and recrystallized grains. Note the progressive increase in the trioctahedral contents as compositions deviate from the muscovite end-member. Higher apparent trioctahedral contents of the pegmatitic type are inferred to be caused by high Fe3+/ FeH ratios. (c) Compositional variation within a representative single primary crystal from sample T316.

cal uncertainty. Because of the heterogeneous nature of individual primary crystals, we have used the complete set of analyses reported in Appendix Table 1, which rep- 100 150 200 resents single spot compositions located within compo- Time (seconds) sitionally homogeneous portions at the scale of BSE images. Compositional changes in the matrix grains (primary Fig. 2. Plots showing the effect of the duration of electron and recrystallized) include a decrease in Si (6.66-6.14 bombardment on measured X-ray intensities of Na and K in atoms pfu), Fe (0.46-0.10), Mg (0.33-0.05), Ti (0.13- micas. The diagrams represent the percent variation in counting 0.01), and an increase in [4]AI (1.34-1.86), [6]Al (3.17- rates (counts/second) during successive 10- and 20-s intervals relative to the initial 10-s integration period. The data were ob- 3.84), Na (0.10-0.20), K (1.62-1.77), and total interlayer tained during single continuous analyses on the same spot of occupancy (1.74-1.94). The octahedral occupancy com- micas from sample T313 referred to in the text, using a constant pares well with other published analyses of muscovite (cf. acceleration potential of 20 kV. The results of experiments using Miller et aI., 1981; Guidotti, 1984) and decreases (4.10- probe current densities of 0.8 nA/~m2 (open symbols) and 20 4.00) as primary muscovite becomes depleted in Fe and nA/~m2 (solid symbols) are shown for comparison. The plots Mg (Fig. 4). The good correlations found among Si, Mg, [6] indicate that no significant alkali mobilization takes place during Fe, Fe + Mg, and Al (Fig. 4, Table 2) indicate that the the initial 50 s when moderate current densities of about 1 nA/ progress of Reaction 1 is responsible for the largest frac- tLm2 are used (e.g., 20 nA with a probe diameter of about 5-6 tion of the observed compositional variations in matrix ~m). For biotite, the same applies up to much higher probe cur- muscovite. On the other hand, the pegmatitic muscovite rent densities, which allowed the use of a probe size of approximately 0.25 tLm2 to analyze small biotite intergrowths within grains form a homogeneous compositional group (Figs. muscovite. 3, 4). Their composition overlaps with recrystallized grains having the lowest Si values (mean 6.13 atoms pfu, Table 1); however, pegmatitic muscovite has a lower alkali con- MINERAL COMPOSITION tent (Fig. 4). Muscovite The presence of Fe3+ in matrix muscovite is implied A continuous compositional spectrum is formed by the by the relatively poor correlations of Fe with Si and [6]Al, primary and recrystallized grains (Figs. 3, 4). The group as compared with those of Mg, and the positive correla- of primary muscovite samples exhibits the widest inter- tion between Fe and total octahedral cations (Fig. 4, Ta- and intragrain compositional range. Recrystallized grains ble 2). Normalizing to 44 negative charges with Fe3+ un- show a much narrower intergrain compositional range, accounted for results in an overestimation of all cations and individual crystals are homogeneous within analyti- in the structural formula, thus increasing the calculated GARCIA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES 163

TABLE1. Representative analyses of muscovite (single spots) and biotite

Biotite Muscovite T336 Matrix Intergrown* T316 T336 T506 Prm* Rcrt* pegm Mean Mean (n 2) Spot 1 Spot 79 Spot 12 (n = 11) u = U Si02 48.96 45.76 46.53 34.432 0.649 34.695 0.146 Ti02 1.27 0.21 0.10 2.867 0.208 3.338 0.067 AI203 28.11 36.03 36.87 19.339 0.462 19.452 0.148 Cr203 0.00 0.03 0.00 0.014 0.016 nd FeO 3.67 1.01 1.49 24.267 0.775 22.614 0.036 MnO 0.06 0.02 0.02 0.148 0.027 0.165 0.004 MgO 1.65 0.26 0.26 4.407 0.203 4.800 0.093 CaO 0.01 0.01 0.00 0.016 0.017 0.010 0.010 Na20 0.42 0.77 0.68 0.195 0.046 0.138 0.033 K20 9.68 10.17 10.13 8.875 0.071 8.691 0.036 Total 93.82 94.27 96.06 94.561 0.335 93.900 0.132 H2O** 4.407 4.472 4.556 3.823 0.019 3.835 0.007 Si 6.661 6.136 6.124 5.401 0.083 5.425 0.013 [4]AI 1.339 1.864 1.876 2.599 0.083 2.575 0.013 [6]AI 3.169 3.830 3.843 0.977 0.039 1.010 0.021 Cr 0.000 0.003 0.000 0.002 0.002 nd Ti 0.130 0.021 0.010 0.338 0.024 0.392 0.007 Fe 0.417 0.113 0.164 3.184 0.113 2.957 0.001 Mn 0.007 0.003 0.002 0.020 0.004 0.022 0.001 Mg 0.334 0.052 0.051 1.030 0.043 1.119 0.020 [6]~ 4.057 4.023 4.069 5.551 0.057 5.501 0.005 Ca 0.001 0.001 0.000 0.003 0.003 0.002 0.002 Na 0.112 0.200 0.174 0.059 0.014 0.042 0.010 K 1.681 1.740 1.700 1.776 0.019 1.734 0.010 [12]~ 1.794 1.942 1.874 1.838 0.013 1.777 0.019

Note: Total Fe expressed as FeO. Cations normalized to 20 0 atoms and 4 OH; n = number of analyses; nd = not determined. Petrographic groups of muscovite: Prm = primary, Rcrt = recrystallized, and pegm = pegmatitic grains (see Appendix Table 1 for the complete set of analyses). Matrix biotite represents an average of high-Ti (>0.3 atoms pfu) compositions (see text), and the intergrown biotite corresponds to the biggest lamellae occurring within the primary muscovite crystal of Figure 5. Analyses used in the mass balance calculations (Eq. 7 in the text). * ** Calculated H20 wt% on the basis of 4 OH pfu.

octahedral occupancies. The recrystallized and pegmatit- 0.5 0.5 0.9 ic grains have a higher octahedral occupancy (mean of o(Jf) 0 0.4 o 0 0.4 4.03 and 4.07, respectively) than expected as a continu- 0.6 0.3 0 .; ation of the trend formed by the primary grains and rel- .:: a t 02" 0.3 ~ ""' ative to their Fe + Mg contents (Figs. 3, 4). This suggests 0.1 that the Fe3+/Fe2+ ratio in these muscovite grains is high- O'~.O 6.2 6.4 6.6 6.8 6.2 6.4 6.6 6.8 er. The presence of Fe3+ might also account for the ratio Si Si Si Fe/Mg > 1 in our samples, which increase from the pri- 0.15 0.9 (J mary (mean 1.178, = 0.236) to the recrystallized (mean o ~o o (J (J 0.10 0.6 1.808, = 0.282) and pegmatitic (mean 3.214, = 0.505) e ~ c§:>bt ::g grains, as compared with the experimentally determined t 0.05 0.3 greater relative solubility ofMg in muscovite (Velde, 1965, ""' 1967; Monier and Robert, 1986a). Any rigorous evalua- o.OIb.o 0.3 0.6 0.9 4.05 4.10 tion of the Fe3+ contents in the muscovite is precluded Fe+Mg 161Sum by the probable occurrence of octahedral occupancy in 1.80 0.20 2.0 excess of 4 atoms pfu (Monier and Robert, 1986a; Mas- 0.18 1.75 1.9 sonne and Schreyer, 1987), as the amount of the calcu- 0.16 S ro ;:0:::1.70 Z a t lated Fe3+ (normalizing to eight tetrahedral plus four oc- 0.14 :g 1.8 tahedral cations and 44 negative charges) is proportional 1.65 0.12

to the octahedral occupancy (normalizing to 22 0). How- 1.6~.0 6.2 6.8 0.1~.0 6.2 6.4 6.6 6.8 1.1>.0 6.2 6.4 6.6 6.8 ever, the formulae that include calculated Fe3+ can be Si Si Si considered as end-member estimates. For matrix mus- Fig. 4. Selected bivariate diagrams showing the composi- covite, these calculations resulted in a slight modification tional spectrum of the entire muscovite data set. Circles pri- of the compositional ranges of the major elements and = mary grains. Squares = recrystallized grains. Triangles = peg- little or no change of the minor elements (compare the matitic grains. The regression lines do not include the pegmatitic [6]AI ranges given above with Si = 6.63-6.12, = 3.82- grains. 164 GARCIA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES

TABLE 2. Pearson correlation matrix of selected variables for primary and recrystallized muscovite

Si [6JAI Ti Fe Mg [6J~ Na K [12J~ Na/K Fe + Mg

Si 1.000 [6JAI -0.932 1.000 Ti 0.813 - 0.888 1.000 Fe 0.765 -0.911 0.674 1.000 Mg 0.975 0.978 0.851 0.854 1.000 [6J~ - 0.225 - 0 .408 0.098 0.715 0.363 1.000 Na -0.838 0.853 - 0.849 -0.671 -0.863 -0.149 1.000 K 0.538 0.321 0.275 - 0.196 0.453 -0.156 0.214 1.000 [12J~ - - - -0.814 0.649 -0.610 - 0.465 -0.758 -0.192 0.630 0.893 1.000 Na/K -0.781 0.826 -0.827 - 0.657 -0.818 -0.130 0.991 0.080 0.519 1.000 Fe + Mg 0.893 - 0.978 0.784 0.970 0.955 0.576 -0.786 -0.322 - 0.619 -0.758 1.000 Note: Number of observations: 61.

3.12, Mg = 0.33-0.05, Ti = 0.13-0.01, Na = 0.20-0.10, MICROTEXTURES and K = 1.76-1.61), indicating that only minor errors Primary and recrystallized muscovite are introduced in the other components by the chosen The compositional heterogeneities within individual normalization, with Fetotas Fe2+. This is not the case with primary muscovite grains are generally associated with regard to the pegmatitic muscovite, whose octahedral occupancy is largely overestimated as a probable conse- the biotite-quartz intergrowths. Halos depleted in Si, Fe, and Mg and enriched in Al develop close to these inter- quence of very high Fe3+IFe2+ratios (see below). growths (Fig. 5), indicating that they were produced by Other minerals phengitic decomposition of primary muscovite, in agreement with the observed bulk compositional changes of Biotite also exhibits strong intergrain heterogeneities (Appendix Table 2). Matrix crystals are Ti-rich (>0.3 matrix muscovite (Fig. 4). The discontinuous and irregular outer rims appear to be related to the formation of atoms pfu), intermediate Ti contents are found in biotite larger biotite grains and overgrowths and to grain-size pseudomorphs (+ Ms + PI) after gamet, and low-Ti «0.1 reduction and recrystallization at the edges of the crystals atoms pfu) crystals are retrograde pseudomorphs after (Fig. 5A), although internally recrystallized grains are also garnet. This indicates that gamet-consuming reactions are apparent (Fig. 5A, 5B). Since recrystallized grains are the responsible for the change in composition of the matrix end product of the main compositional trend of primary biotite crystals. The decrease in Ti involves a parallel crystals bearing Bt + Qtz intergrowths (Figs. 3, 4), their increase in [6]Al,Fe, and octahedral occupancy and a de- compositions can be directly related to the same contin- crease in Mg. On the other hand, the composition of in- uous reaction affecting the primary muscovite. In the fol- tergrown biotite within primary and pegmatitic musco- lowing paragraphs microtextural observations related to vite is. relatively Ti-rich (0.39 and 0.25 atoms pfu, respectIvely, Table 1 and Appendix Table 2). Gamet is the mechanism of generation of the biotite-quartz intergrowths are summarized, shedding light on the kinetics normally ~nzoned, being rich in Fe and Ca (xGrs > 0.2) of primary muscovite decomposition. and poor In Mg and Mn (xsps < 0.03, Appendix Table 2). Heterogeneous nucleation. A variety of defects and de- Lower Ca and higher Mn contents are recorded in some fective areas appear over broad areas of primary mus- fine-grained crystals and in thin rims (up to 50 ~m) of the covite grains. Most of these are deformational structures coarser grains. This zonation is crosscut by the Ms + Bt resulting from the shear deformation suffered by these PI pseudomorphs. Plagioclase is rich in albite (Appen- rocks, and they include abundant stacking faults, elastic dIX~ Table 2). The compositional variations within indi- bending of the layers, edge dislocations (Fig. 6), and local :idual. samples are related to several stages of growth, pull-apart microstructures (Fig. 7). Because the stored InvolvIng a progressive increase in the anorthite content. elastic and surface energy reduces the activation energy In sample T313, crystals included in Kfs megacrysts ap- required for the formation of a nucleus of critical size pear with a slight reverse zoning with xenomorphic cores defects are favorable nucleation sites (e.g., Putnis and (xAn= 0.09) and idiom orphic overgrowths (xAn= 0.13), McConnell, 1980; Lasaga, 1981; Ridley and Thompson, 0.14). w~er~as matrix grains are homogeneous (xAn = 1986). In our images, this is reflected by the abundance WIthIn the pseudomorphs after gamet, plagioclase bears higher anorthite contents (XAn= 0.19). Potassium feldspar of small intergrowths (Le.,. abundant nucleation sites). A also has minor compositional intergrain heterogeneities particular case was found in relation with pull-apart mi- w~thin individual samples (Appendix Table 2), being crostructures. These microstructures formed as voids and shghtly less albitic in the fine-grained matrix grains (XAb mic~ocracks along (001) planes (Fig. 7), probably along = 0.19-0.18), as opposed to the larger grains (XAb= 0.23- the Interlayer region, which contains large and less strongly 0.21), except when the latter appear slightly exsolved (e.g., bonded cations (Veblen, 1983). In the example illustrated T313, XAb= 0.18). in Figure 7, the plastic strain is stored as two opposite GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES 165

Fig. 5. BSE images and qualitative element traverses from a (C), and (D) Enlarged area marked by arrow in A showing Fe primary muscovite crystal from sample T336 bearing abundant and Mg depletion halos (dark gray) in muscovite (gray) close to Bt + Qtz intergrowths. (A) Low-magnification view of the crys- intergrown biotite (white), and a tabular dark recrystallized mus- tal. The abundant biotite lamellae (bright white zones) and quartz covite grain. (E) and (F) Enlarged area marked by arrow in A inclusions (dark rods) parallel to the (001) planes of muscovite showing elongated rods of quartz (black) and biotite laths (white) stand out clearly. Muscovite appears as various shades of gray. and the depletion in Si and enrichment in Al of muscovite (gray) Note the irregular outer rim, and the patchy diffusion zoning close to the intergrowths. The qualitative element traverses are within muscovite. Recrystallized muscovite grains appear as high- expressed in counts per second. angle inclusions and at the edges of the primary muscovite. (B), 166 GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES

Fig. 7. Shear pull-apart microstructure parallel to the muscovite layers. Two edge dislocations within muscovite disrupt the structure that probably grew through a climb process. The two poorly defined extra fringes that appear attached to these coupled edge dislocations might represent single atomic rows of a growing phase (biotite?) in the void or Fresnel fringes at the edges of the mica along these gaps. Fig. 6. TEM images of a composite intergrowth, within a primary muscovite host, consisting of two 2oo-A biotite lamellae separated by an inner low-contrast (quartz?) band. (A) Low- The coincidence of the c* axes of muscovite and very magnification electron micrograph of this threefold microstruc- thin biotite lamellae is indicated by lattice fringes with ture. (B) An enlarged image of the region outlined in A showing 10-A spacing in both phases (Fig. 8). Since Figure 8 is abundant defective areas (thin arrows) consisting of elastic bend- not a structure image, no direct information can be gained ing of the layers and edge dislocations in muscovite. Biotite la- relative to the structural continuity along other direc- mellae show only minor deformation as compared with mus- tions. However, the lack of boundaries with sharp con- covite. The semicoherent muscovite-biotite interface is indicated trast, such as those related to stacking faults of micas or by the thicker arrows: no change in orientation along a and b other phyllosilicates (e.g., Veblen, 1983, his Fig. 2), which directions of the micas across the interface is inferred from the would indicate a change in orientation in a and b direc- absence of marked contrast. tions of the lamellae (D. R. Veblen, personal communi- cation, 1991), suggests the development of semicoherent edge dislocations, so that the initial continuity between (001) interfaces for this very thin lamella (i.e., during the the two muscovite layers affected by the pulling apart is initial stages of growth). actually disrupted. In this figure, two poorly defined sin- The above structural relations suggest that (001) planes gle fringes (the arrowed darker fringe inside the lighter might have acted as optimal phase boundaries (Robinson area of low electron density) appear attached to the two et aI., 1971) for both micas, in spite of the fact that their edge dislocations disrupting muscovite. These extra fringes different a and b dimensions would make an exact fit of could be single atomic rows of a quenched c\ngstrom- both structures along (001) planes impossible (see lijima sized phase (biotite?) growing in the void as the micro- and Zhu, 1982, for coherent biotite-muscovite inter- structure migrates through the climb of the edge dislo- growths perpendicular to mica layers). The a and b di- cations. An alternative interpretation is that these apparent mensions are related to the coplanarity of the basal 0 of extra fringes do not actually result from an extra layef, the tetrahedral layers and the geometry of the octahedral but rather from the Fresnel fringes at the edges of the coordination in micas and are a function of the octahe- mica along these gaps (see also Eggleton and Buseck, dral cations and occupancy (Zussman, 1979; Weiss et aI., 1980). 1985). Thus, coherence was favored during initial stages Semi coherent growth. TEM images further confirm the of growth by the higher phengite contents of the host evidence for structurally controlled (interface-controlled) muscovite (i.e., by its greater a and b dimensions; Zuss- growth of biotite, and to a lesser extent quartz, parallel man, 1979; Massonne and Schreyer, 1986), although it to (001) planes of the micas. However, variable degrees was not maintained with further decomposition as the a of apparent coherency relative to the c* direction (Figs. and b dimensions of muscovite shrink. With continued 6, 8, 9) were found to be related to the grain size of the decomposition and biotite growth, the elastic strain fields biotite lamellae, and thus to different stages of growth. at the semicohefent interfaces would be expected to be- GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES 167

Fig. 8. Semicoherent biotite lamella (dark) within muscovite. Fig. 9. Low magnification TEM image of intergrown biotite Continuity of a and b directions in both micas across the inter- lamellae of various sizes. The thicker one (bottom) bears irreg- faces is inferred from the absence of sharp boundaries between ular boundaries and is surrounded by quartz rims. the two phases. come too large and would lead to the development of as is indicated by diffusion halos around the intergrown defects and increasingly incoherent interfaces. This al- phases (Fig. 5). Regardless of being an unfavorable dif- lows a reduction of the strain energy but increases the fusion pathway (e.g., Fortier and Giletti, 1991), this pro- surface energy (Putnis and McConnell, 1980) and is il- cess could have been favored because of the minimiza- lustrated by the bending of the layers close to the inter- tion of the interfacial energy, at least while the migrating faces of relatively small lamellae (Fig. 6). Figure 6A shows phase boundaries maintained crystallographic semico- a threefold structure consisting of an inner low contrast herency. The development of incoherent boundaries and band (quartz?) surrounded by two biotite lamellae (each wide depletion halos in the host muscovite must have led about 200 A thick) included in muscovite. Abundant edge to a change from the initial surface-controlled semicoher- dislocations involving (001) planes in muscovite (Fig. 6B) ent growth to a growth controlled by volume diffusion. indicate the lack of perfectly coherent boundaries, al- The presence of thick biotite lamellae rimmed by quartz though the dislocations might in fact be an artifact of the (Fig. 9) suggests mass transfer by diffusion of dissolved bending of the layers causing apparent shifting of the species in solution through the grain interfaces and growth fringes (Guthrie and Veblen, 1989, 1990). As noted pre- by precipitation from a fluid. viously, the orientations of the a and b directions of ad- The above observations indicate that the progress of jacent muscovite-biotite pairs in this figure also appear the decomposition reaction at a particular site was con- to be the same across the two interfaces. The symmetry trolled by the rate of volume diffusion, although a com- of this structure suggests that nucleation took place in a plex picture of the reaction pathway is envisaged from planar structure (stacking fault?) away from which the the large number of intergrowths (i.e., nucleation sites), growth of biotite had taken place and where quartz is now their wide range of sizes and degree of coherency (i.e., located. Since the biotite layers bear only minor defor- stages of growth), and the role of the continuous hetero- mation compared with the host muscovite, their nucle- geneous deformation and recrystallization accompanying ation on defective areas of this type and the accommo- muscovite decomposition. In any case, the close spatial dation of both phases appear to have benefited from the relationship between the intergrowths and the diffusion already deformed structure of muscovite. The develop- halos in muscovite, the fine-grained nature of the inter- ment of semicoherent to incoherent interfaces is hence growths, and their heterogeneous distribution within the inferred to have resulted from both nucleation and growth core of the primary muscovite crystals (Fig. 5) suggest on defective areas and the early stages of the coarsening that garnet decomposition was not directly related to the of previously semicoherent biotite intergrowths. growth of the intergrown phases. Coarsening. The coincidence of the c* axes is lost, and low angle grain boundaries develop, when biotite lamel- Pegmatitic muscovite breakdown lae get thicker (more than about 500 A). The advance of The reaction textures associated with the pegmatitic replacement across the (001) planes of the muscovite grains are also distinctive (Fig. 10). Fibrolitic sillimanite would imply that the diffusion of decomposing products is associated with potassium feldspar in thin reaction rims proceeded approximately perpendicular to mica layers, around the muscovite crystals (Fig. lOB), whereas coarse

~-- 168 GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES

Fig. 10. Optical and BSE images of muscovite breakdown ite + potassium feldspar envelope (plane-polarized light). (C) textures within a pegmatitic crystal from sample T506. (A) Op- and (D) BSE images of the potassium feldspar + andalusite tical image of the chessboard intergrowths of potassium feldspar chessboard spot shown in A. Skeletal andalusite (black) appears (black and dark gray) and andalusite (gray and light gray) located as blocky areas intergrown with potassium feldspar (light gray) in the core of the pegmatitic muscovite crystal (white). Partial and biotite (white) in the central area of the spot (C). Toward optical continuity is indicated by simultaneous extinction of dif- the rims (D), biotite and potassium feldspar lamellae are orient- ferent potassium feldspar grains (crossed polarizers). (B) Optical ed parallel to the (001) planes of the host muscovite, causing the image of fibrous sillimanite (fur) and potassium feldspar around ragged appearance of the contact between the reactant muscovite the periphery of a pegmatitic muscovite crystal in contact with and the product assemblage. Note the presence within the prod- quartz; a wider spot of intergrown andalusite and potassium uct assemblage of irregular patches of unreacted muscovite (darker feldspar (lower right comer) is unrelated to the fibrous silliman- gray) associated with andalusite (thicker white arrows).

andalusite and chessboard intergrowths of And + Kfs id-rich environment to facilitate the transport of Si with- develop towards the cores of the crystals (Fig. lOB, 10C). in the pegmatitic crystals; otherwise, these cores of mus- The chessboard intergrowths are rimmed by potassium covite (considered as a Si02-subsaturated subsystem) feldspar and scarce biotite growing parallel to the (001) should not have reacted to produce andalusite. planes of the host muscovite (Fig. 10C, 10D). The prod- Pegmatitic muscovite lacks indications of decomposi- uct assemblages and textural analysis indicate that this tion processes analogous to those already described for muscovite reacted through the breakdown Reaction 2b, primary muscovite (Le., through Reaction I), probably proceeding initially at the grain contacts between mus- due to its limited compositional deviations from pure covite and quartz to produce small amounts of Kfs + muscovite. BSE inspection and qualitative traverses fibrous sillimanite, which partly isolated the cores of showed no compositional heterogeneities in the musco- muscovite crystals from the matrix. The fact that the most vite that borders the product phases, indicating that sig- extensive breakdown took place inside the muscovite nificant continuous net exchanges were not involved and crystals, away from the surrounding quartz, implies a flu- hence supporting the discontinuous nature of Reaction GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES 169

2b (cf. A.B. Thompson, 1982). What is notable is that (1) these muscovite samples did not react to completion, and (2) the matrix muscovite crystals lack any textural evidence for breakdown through Reaction 2b. These facts constrain the interpretation of these reaction textures in terms of the amount of overstepping required for the formation of critically sized nuclei (see below).

P-T-t EVOLUTION 600 700 800 Gamet-biotite (GARB) thermometry (Thompson, 1976; T (C) Ferry and Spear, 1978; Hodges and Spear, 1982; Perchuk and Larent'eva, 1983; Ganguly and Saxena, 1984; In- dares and Martignole, 1985; Berman, 1990) and gamet- plagioclase-biotite-muscovite (GPBM) barometry (Ghent and Stout, 1981; Hodges and Crowley, 1985; Hoisch, 1990) were selected for P- T estimations in the analyzed banded gneiss (Fig. 11). Difficulties in the application of both equilibria to these rocks arise from the compositional heterogeneities displayed by the intervening phases, such that conditions of equilibrium cannot be demon- T (C) strated. The compositions used, considered to approach early equilibration conditions, are the homogeneous high- Fig. 11. (a) P-T diagram showing selected mineral and melting reactions within the KNaFMASH (dashed curves) and Ca garnet core, high- Ti matrix biotite, high-Si primary KFMASH (solid curves) systems. All reactions after A.B. muscovite, and matrix plagioclase (Table 1). Thompson (1982), except the Si isopleths for Reaction 1 in the A large range of temperatures (600-750 °C) are ob- text (numbers without boxes) that are after Massonne and tained when different calibrations of the GARB Fe-Mg Schreyer (1987) and the aluminum silicate transitions (after Hol- exchange equilibrium and activity expressions are ap- daway, 1971). The remaining reactions are labeled following the plied to the analyzed samples. Figure 11 presents the end- numbers of A.B. Thompson (1982, his Fig. 7, written with the member T estimates, together with error bands (20) based low-T assemblage to the left) as follows: 33': Fe-Ms + Ab + Fe- on analytical error propagation. As a general result, the Bt + Ksp + Qtz + V = L, 5': Fe-Ms + Ab + Qtz = Kfs + Als highest temperatures are obtained when garnet activity + Fe-Bt + V, 34': Kfs + Ab + Qtz + Fe-Bt + V = L, 37': Fe- models (Hodges and Spear, 1982; Ganguly and Saxena, Ms + Ab + Qtz = Fe-Bt + Kfs + Als + L, where Als = AI2SiOs, 1984; Berman, 1990) are included to correct the experi- V = H20, and L = liquid; Reactions 33 and 5 (KFMASH system) are similar to 33' and 5' (KNaFMASH system), respec- mental calibration of Ferry and Spear (1978). Composi- tively, but excluding Ab. Reaction 5 in this figure is Reaction tional corrections to account for deviations of biotite from 2b in the text. The dotted arrow indicates the path followed by the pure Mg-Fe system (Indares and Martignole, 1985) the studied rocks, inferred from thermo barometric estimates result in lower temperature estimates (in the range of 600- (below), the compositional range of muscovite, and the break- 650 °C), whose error bands overlap with solutions of both down product of pegmatitic grains. (b) (c), and (d) P-T plots of the experimental calibrations (Ferry and Spear, 1978; equilibria for three of the samples studied. The three curves Perchuk and Larent' eva, 1983), and the empirical cali- shown for each biotite + garnet equilibrium (dashed curves la- brations (Thompson, 1976). These latter estimates are beled I&M = Indares and Martignole (1985), model A; solid considered to approach more closely the early equilibra- curves labeled H&S = Hodges and Spear (1982) correspond to tion conditions. The empirical calibrations of the GPBM the calculated curve for the selected compositions (central line) equilibrium of Hodges and Crowley (1985) and Hoisch and to 20"error bands, based on analytical error propagation on Kd[=(Mg/Fe)Grt/(Mg/Fe)Bt]. The curves shown for garnet + pla- (1990, his Reaction R6) yield pressure estimates in the gioclase + biotite + muscovite equilibria (labeled GPBM, Hodges range of 12-14 and 13.5-17 kbar, respectively (calculat- and Crowley, 1985) represent the selected compositions with ed at 650°C; the molar volume of grossular calculated line types, according to plagioclase type as follows: solid curve following Newton and Haselton, 1981). These estimates = matrix; dashed and dotted curves (sample T313) = rim and are considered to have low accuracy because the com- core, respectively, of plagioclase inclusion within potassium positions used by these authors in their regression cal- feldspar phenocryst. In all plots, the Si isopleth of Reaction 1 culations differ significantly from those present in the an- corresponding to Si = 6.6 atoms pfu is shown for reference. alyzed banded gneisses, particularly in the case of the micas (note that Hodges and Crowley, 1985; and Hoisch, 1990, used pelitic assemblages bearing any of the Al2SiOs mum precision of :t2 kbar and :t 100 °C for a given polymorphs, so that the micas are saturated in AI). In calculation). Because of these problems of precision and addition, a large uncertainty arises from the possible in- accuracy, no proper error propagation calculations were accuracies of other equilibria used for the regression cal- attempted, and Figure 11 presents the results using the ibrations (Hodges and Crowley, 1985, suggested a maxi- calibration of Hodges and Crowley (1985). These esti-

~"-_._- -- --~.._._---

170 GARCIA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES mates do nonetheless indicate pressures greater than 10 and unpublished data). Continued rapid decompression kbar, in agreement with the Si content of matrix mus- is suggested by the deposi tiona! age of 19 :t: 1 Ma for the covite. The graphical celadonite barometer of Massonne nappe-sealing sediments of the La Vifiuela formation and Schreyer (1987, as represented in our Fig. lla) yields (Gonzalez-Donoso et aI., 1982), which includes boulders ca. 11 kbar (at 650°C) with the maximum Si content of of medium-grade graphite schists belonging to the Torrox the analyzed primary muscovite (6.66 atoms pfu). unit (see Fig. 1). These data indicate very high rates of After these high-pressure initial conditions, decom- cooling (100-300 °C/m.y.) and uplift (2.5 to > 5 km/m.y.) pression and lower T conditions are indicated by the low- for the late low-P stage of the decompression path (Zeck er Ca and higher Mn contents of garnet rims, either if et aI., 1989, 1992; Monie et aI., 1991a). these are considered as a growth characteristic (cf. Green, 1977) or a diffusion-induced feature at relatively high T, DISCUSSION: MODEL DECOMPOSITION AND and by the PI-Bt-Ms pseudomorphs after gamet, which BREAKDOWN REACTIONS crosscut this zonation. The compositional heterogeneities and reaction textures of muscovite also indicate that rel- Phengitic decomposition atively large changes in P affected these rocks since the Qualitative reaction pathway. One significant feature of early crystallization and equilibration of primary mus- the biotite-quartz intergrowths within primary muscovite covite. The decrease in Si content (6.66-6.14 atoms pfu) is the lack of a distinct minimum size (i.e., 10-6-10-2 mm accompanying the decomposition and recrystallization of along c* for the case of biotite), indicating that nucleation matrix muscovite would indicate a change in pressure did not end as the reaction progressed (see Ridley, 1985). from ca. 11 to 2-3 kbar (Massonne and Schreyer, 1987), The fact that new nucleii were continuously generated whereas the breakdown of pegmatitic muscovite to And during decomposition of primary muscovite is explained + Kfs + Bt pseudomorphs through Reaction 2b indicates by (1) the continuous nature of the decomposition reac- that the rocks reached low pressures at relatively high tion, (2) a high decompression rate at intermediate to temperatures, Le., ca. 600 °C and 2 kbar (Chatterjee and high pressure, and (3) a change from surface- to diffusion- Johannes, 1974; A.B. Thompson, 1982) (Fig. 11). controlled growth of the thicker intergrowth with the Although large errors may be associated with the cal- progress of reaction. Textural analysis of the reactant pri- culated pressures and temperatures, we infer roughly 10 mary muscovite and product biotite and quartz indicates kbar of decompression with minor cooling (50-100°C) that the development of depletion halos within musco- (Fig. 11a). The presence of both deformed and crosscut- vite close to the biotite-quartz intergrowths hampered ting aplite bodies indicates that continuous ductile defor- further growth and even caused the cessation of decom- mation was affecting the Torrox gneiss complex while position at a particular growing site when the halos had partially molten, Le., that the main deformational event reached a critical width (in the range of 5-1 0 ~m). The started at high P-T and proceeded to subsolidus condi- large free energy changes of reactions involving the relict tions. The processes of decomposition and recrystalliza- high-Si compositions are proportional to the degree of tion affecting the primary muscovite also suggest that de- overstepped conditions (Ridley and Thompson, 1986) and formation was taking place during decompression, as should have preserved and even increased the nucleation recorded by the nucleation of biotite in deformation de- rate. This favored the progress of reaction in these not- fects. This close relationship between deformation and yet-decomposed areas by further nucleation (either on near isothermal decompression implies that the main de- continuously generated deformation defects or at ran- formation recorded in the TGC is associated with crustal dom) and semicoherent growth of biotite. At the high extension rather than thickening (e.g., Thompson and temperature of decomposition, this reaction pathway ne- Ridley, 1987), opposite to the interpretation of Cuevas et cessitates a high decompression rate; otherwise further ai. (1989), who associate it with nappe emplacement. In- growth of previously formed intergrowths would have deed, the occurrence of rapid synmetamorphic extension- been favored rather than an extremely high number of al tectonics associated with the main deformation event intergrowths. This resulted in the heterogeneous spatial is consistent with the clustering of cooling ages of several distribution and sizes of the product phase and of the isotopic chronometers (Rb/Sr and 4°Ar/39Ar in micas and diffusion-induced zoning in the reactant muscovite. feldspars) around 21 :t: 2 Ma in rocks from the TGC and Under these circumstances, decomposition must have adjacent metapelites (Zeck et aI., 1989, 1992; Monie et proceeded irreversibly as the relict composition of mus- aI., 1991 a). As an example, specimen T3 37, a porphyritic covite in these not-yet-decomposed areas was increasing- muscovite biotite gneiss not presented in this paper that ly displaced from equilibrium during decompression (cf. also contains Si-rich primary and low-Si recrystallized Ridley and Thompson, 1986, their Fig. 4). The lack of muscovite, yields cooling ages of 22.4 :t: 0.7 Ma (Rb/Sr quantitative compositional data on the finer biotite in- muscovite-whole rock), 19.0 :t 0.7 Ma (4°Ar/39Ar mus- tergrowths precludes the evaluation of a hypothetical covite), 20.3 :t 0.3 (4°Ar/39Ar biotite), and 20.0 :t 0.6 crystallization of biotite with a disequilibrium composi- (low- T 40Ar /39Ar potassium feldspar), and the matrix tion; however, the absence of conspicuous potassium muscovite from sample T316 yields a 4°Ar/39Ar age of feldspar within the intergrown assemblage could be con- 19.4 :t: 0.4 (Zeck et aI., 1989, 1992; Monie et aI., 1991a, sidered an indication of a metastable decomposition pro- GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES 171

TABLE3. Components and net-transfer reactions (KTMTiASH 0.50 system) describing the decomposition of primary muscovite 0.25 Phase components Additive components Exchange components 0.00 Muscovite (Ms) K2AI6Si602o(OH)4 (Ms) Tk, Prl, di-tri, Ti-Spl 0.25 Biotite (Bt) K2Mg6AI2Si602o(OH)4 (Phi) Tk, Prl, di-tri, Ti 0 Potassium 0.15 feldspar (Kfs) KAISi30a (Or) Quartz (Qtz) Si02 (Qtz) 0.05 Rutile (Rt) Ti02 (Rt) Fluid H20 -0.05 Net-transfer reactions (mole units) 0.08 3 Ms + 6 Tk = 4 Or + 6 Qtz + Phi + 4H20 (1) (3) 6 Or + 6 Prl = 24 Qtz (4) 3 Ms + 6 di-tri = 3 Phi 0.04 (5) 3 Ms + 6 Ti-Spl = 3 Phi + 6 Ti 0 (6) Ti -rns 3 Ms + 6 Ti-Spl = Phi + 6 Rt + 4 Or + 4H20 0.00 Note: Exchange component abbreviations (reference is made to selected O~5 [4JSi[6JMg[4JAL1[6JAL1 experimental investigations): Tk = (Tschermak's substitution; Velde, 1965, 1967; Monier and Robert, 1986a; Massonne and 0 [6JMg[6JTi[6JAL2(named titanium spinel as in Dy- o~o 0 Schreyer, 1987), Ti-Spl = mek, 1983), Prl [4JSi[12JD[12JK_1[4JAL1(pyrophyllite; Velde, 1969; Rosen- o~ ~ = O~5 .000 berg, 1987); di-tri [6JMg_3[6JAL2[6JD_1 = (di-trioctahedral; Monier and Robert, prl ° ° ~ Ti[6JDMg_2 (Ti D; Forbes 1986a; Massonne and Schreyer, 1987), Ti D = 0.00 ~~ .. and Flower, 1974; Abrecht and Hewitt, 1988). 0.06 o o cess, since Reaction 1 predicts detectable amounts of this 0.03 phase. Nonetheless, as will be shown below, we interpret this major divergence among the expected and observed 0.00 phase assemblage produced by the decomposition of pri- 0.12 mary muscovite in terms of (1) the effect of other com- pa o ponents of muscovite in the final product assemblage, 0.08 and (2) diffusion of K to the matrix. Mass balance calculations and net-transfer reactions. 0.04 The bulk decomposition reaction can be evaluated fol- 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 lowing the algebraic approach of J.B. Thompson (1982a, IDS 1982b), in terms of a set of independent net-transfer re- Fig. 12. Bivariate plots illustrating the compositional varia- actions among independently variable components de- tions within the investigated muscovite populations (symbols as fined as linearly independent exchange vectors and ad- in Figs. 3, 4). The regression lines do not include the pegmatitic ditive components (Table 3). For the solid solutions of crystals. The seven-component cation basis (KNaFMTiAS) has muscovite and biotite, with muscovite and phlogopite, been transformed into the molecular basis resulting from the respectively, as the additive components, the number of operation of the exchange components mentioned in the text and Table 3 on the muscovite additive component (see Greenwood, independent exchange vectors to be defined is six, since 1975; J.B. Thompson, 1982a, for details on the procedure). In the number of stoichiometric and charge-balance con- addition to muscovite, the ensuing molecular species are leuco- straints that must be satisfied is four (Le., total positive phyllite [lcp, K2Mg2AI2Sig02o(OH)4], titanium muscovite [Ti- [4]Al charge = total negative charge = 44, Si + = 8, [6]AI ms, K2Mg2Ti2AI2Si602o(OH)4], pyrophyllite [prl, Al4Sig02o(OH)4], [6]0 [12]0 + Ti + Fe2+ + Mg + = 6, and K + Na + = 2), phlogopite (phI, K2Mg6AI2Si602o(OH)4], and paragonite (pa, and the number of compositional variables in the phases Na2AI6Si602o(OH)4]. Negative values of the leucophyllite com- is ten (cf. Labotka, 1983; Hewitt and Abrecht, 1986). For ponent are obtained in the case of the pegmatitic grains, where muscovite, in addition to the exchange vectors fm the estimated trioctahedral and Ti components account for more (FeMg_1) and nk (NaK_1), the exchange components Mg + Fe than its actual value. In this case, higher Fe3+IFe2+ shown in Table 3 were selected to account for the ob- ratios may have led to an overestimation of the octahedral occupancy when total Fe is expressed as Fe2+. served correlations among the components (Fig. 4, Table 2), and following the suggestions by Hewitt and Abrecht (1986). For biotite, the same exchange components were actions (J.B. Thompson, 1982b), those involving the ex- selected, except for the titanium spinel exchange, which change components fm and nk can be excluded from con- has been substituted for the Ti-vacancy exchange (Table sideration by condensation of the KNaFMTiASH system 3) because the amount of Ti clearly controls the amount into the KMTiASH system (i.e., projecting from FeMg_1 of octahedral vacancies (see Appendix Table 2). and NaK_1). This procedure is warranted by the fact that Since exchange reactions do not substantially modify the doublets Fe-Mg and Na-K behaved similarly upon the modal abundances of the phases involved in the re- decomposition of primary muscovite (Figs. 4, 12). The

-- 172 GARCIA-CASCO ET AL.: MUSCOVITE IN HIGH P-TGNEISSES five independent net-transfer reactions (Reactions 1, 3, ing stoichiometric coefficients of the exchange vectors on 4,5, 6, Table 3) relating the 11 selected components with- the left side of Equation 8 (note that the Ti -Spl exchange in the KMTiASH system can be used to describe the has not been split into its corresponding values involved changes in muscovite composition and modal propor- in Equations 5 and 6, but the correction factors for these tions of the product phases, but only to the extent that equations equal %of the stoichiometric coefficients of the the selected exchange components describe composition- Ti-D and Rt components, respectively). The normalized al changes not biased by the normalization procedure. It values of these coefficients in Equation 8 are 63.3°/0 Tk, must be noted that, in practice, the selection of a definite 15.8°/0 Prl, 7.0°10di-tri, and 13.9°/0 Ti-Spl, indicating that set of independent components is not a simple task for the phengitic muscovite decomposition in Reaction 1 ac- complex solid solutions because their number and nature counts for most of the product phases and changes in depend on (1) the degree of completeness of the analyses composition of the matrix muscovite. However, the (e.g., the availability of independent determinations of changes in Ti and the trioctahedral components in mus- Fe3+, H, and 0), (2) the type of normalization used in the covite through Reactions 5 and 4, respectively, contrib- structural formulae (e.g., 22 0 atoms vs. 8 tetrahedral uted mostly in the production of biotite, and the change plus 4 octahedral cations in muscovite), and (3) particular in the alkali content of muscovite, monitored by the py- crystal-chemical choices (e.g., [6]Tivs. [4]Ti).Additionally, rophyllite exchange through Reaction 3, contributed to once a given composition has been normalized, consid- the production of more quartz and less potassium feld- erable freedom exists in the selection given the large spar. This is illustrated in the condensed AKM projection number of potential exchange vectors (Hewitt and of Figure 13, where the observed compositional trend is Abrecht, 1986). These limitations make the selected set located in the middle of the trends anticipated for each of components nonunique, and some of the exchanges of the four exchange vectors operating in muscovite (or might be nonexistent (see below), but it is considered use- five net-transfer reactions). Since the composition ofpri- ful to model the compositional changes in muscovite dur- mary muscovite undergoing decomposition (assemblage ing phengitic decomposition. A in Fig. 13) appears displaced towards the AI-Phl-Ms Mass balance calculations were done by transforming join of the AI-Phl-Ms-Kfs triangle, lower amounts ofKfs the high-Si primary muscovite cation composition (Prm, would have been produced (ca. 25°/0 of the product solid Spot 1 in Table 1) into the coordinate system defined by phases vs. 47°/0 predicted by Reaction 1, in oxy-equiva- recrystallized muscovite (Rcrt, Spot 79 in Table 1), in- lent units). This is largely the result of the increasing al- tergrown biotite (mean in Table 1), and stoichiometric kali contents in matrix muscovite as decomposition pro- quartz, potassium feldspar, rutile, and H20. The com- ceeds (Figs. 4, 12). positions were recast into the condensed KMTiASH sys- The above results are probably inaccurate because they tem and expressed in terms of oxy-equivalent units (24 depend on the value of the selected set of exchange vec- o units) to approximate modal abundances (cf. Brady tors, which, in turn, critically depends on the value of the and Stout, 1980; Thompson, 1982a). The resulting normalization procedure and the lack of estimates of in- equation dependent Fe3+and H. Particularly important in this con- text is the lack of H estimates and the existence of the Ms (Prm) = 0.647 Ms (Rcrt) + 0.158 Bt pyrophyllite component in muscovite. As long as the in- + 0.089 Qtz + 0.086 Kfs terlayer cations correlated with other compositional vari- + 0.005 Rt + 0.016 H20 (7) ables in the studied muscovite samples (Table 2), care predicts detectable amounts of product potassium feld- was taken to avoid alkali loss during the microprobe anal- spar (Le., the major solid product assemblage would be yses (Fig. 2); other elements, such as Ba, Sr, or Cs, were formed by 47°10Bt, 27°/0 Qtz, and 26°/0 Kfs). As the pro- found to be below the normal quantitation limits (i.e., in portions of phases expected from Reaction 1, in oxy- amounts less than about 0.01 wtO/o),and the calculated equivalent units, are 35°/0Bt, 18°/0Qtz, and 47°10Kfs, the A-site deficiency (ranging from 0.256 to 0.058 atoms pfu) net effect of other components of muscovite upon decom- must be explained as vacancies (Velde, 1969; Rosenberg, position has been a decrease ofKfs and an increase in Bt 1987) or the possible occurrence of H30+ or H20 in the and Qtz, whether or not subsequent diffusion modified interlayer sites (Le., the hydronium substitution H30+IC1; the predicted values. Recasting the compositions in Re- see Dyar et aI., 1991, and references therein). The nega- action 7 in terms of the components shown in Table 3 tive correlation of interlayer cations and Si (Fig. 4, Table yields (in mole units) 2) would favor the occurrence of actual vacancies (i.e., the pyrophyllite substitution) but only to the extent that 3 Ms + 4.510 Tk + 1.130 Prl the hydronium contents had not been sensitive to the + 0.500 di-tri + 0.990 Ti Spl changes in P-T that triggered muscovite decomposition. = 1.342 PhI + 0.527 Ti 0 + 2.185 Or This cannot be ruled out, given the evidence for Hand + 9.031 Qtz + 0.463 Rt + 3.315 H20. (8) o variations in biotite with metamorphic grade and as- Reaction 8 can be considered as a weighted bulk decom- semblage as reported by Dyar et ai. (1991). It is worth- position reaction made up of the simpler net-transfer Re- while to note that the pegmatitic muscovite, believed to actions 1, 3, 4, 5, and 6 corrected by %of the correspond- have formed close to the H20-saturated granite solidus, GARCIA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES 173 bears low alkali contents as compared with those of the recrystallized matrix muscovite, suggesting higher H contents in the interlayer sites. In fact, if H30+ or H20 fill the interlayer vacancies, reactions of the type 9 Or + 6H30+K_1 [Ms] = 1.5 Ms + 18 Qtz + Quartz + 6H20 (9) + Rutile can also explain both the progressive increase of alkali + H20 cations in matrix muscovite with the progress of decomposition and a decrease in the evolved potassium feldspar components relative to that expected through Reaction 1. Thus, although we have followed other workers (e.g., Wang and Banno, 1987) in modeling the A-site changes through the pyrophyllite substitution due to the lack of independent estimations ofH, the calculated pyrophyllite component (Fig. 12) and the relative contribution of Equation 3 to the bulk decomposition Reaction 8 must AI-phI be seen at best as a composite estimate of the effect of both the pyrophyllite and the hydronium substitutions. phI Nonetheless, the compositional data presented confirm lcp,Ti-ms that variations in the interlayer sites attended the phen- (K,Na)AI02 (Fe,Mg)O gitic decomposition of matrix muscovite. The lack of potassium feldspar occurring as inter- Fig. 13. Condensed AKM plot of analyses of primary (cir- growths within primary muscovite grains contradicts the cles) and recrystallized (squares) muscovite, high- Ti (Ti > 0.3 predictions of phase equilibrium considerations and the atoms pfu; circle) matrix, and intergrown (star) biotite (Table 1), mass balance calculations (Eqs. 7, 8). The preferred ex- projected through Si02, Ti02, and H20, and showing the sche- planation is that K had diffused to the groundmass, the matic phase relations (KMASH system) relevant to primary more so when an aqueous fluid phase evolved as decom- muscovite decomposition [abbreviations as in Fig. 12, except AI-phI: aluminum phlogopite, K2[6JAIIMgsAI3Sis02o(OH)4]. The position proceeded. Diffusion ofK through primary mus- three-phase field A, defined by potassium feldspar, muscovite covite crystals can be rationalized as a hydrated com- bearing the maximum Si content (6.66 atoms pfu), and high pound dissolved in the fluid, implying that hydrolysis titanium matrix biotite represents the early equilibration assem- reactions like blage. As decomposition of primary muscovite proceeds, the three-phase field expands (arrow), as represented by the three- 6 Or + 4H20 = Ms + 12 Qtz + 4KOH (fluid) (10) phase field B defined by potassium feldspar, low-Si primary actually represented steps of the molecular reactions. In muscovite and intergrown biotite. The compositional deviations this regard, the increase in the alkali content of primary of muscovite undergoing decomposition from the ms-Icp join muscovite as decomposition proceeds can be envisaged toward pyrophyllite and phlogopite end-members would pro- to have taken place through exchange with the K-bearing duce less potassium feldspar upon decomposition relative to the amounts predicted by Reaction 1 (note also that the deviations fluid through the simple reaction of muscovite due to the Tk and Ti-Spl exchanges are collinear KOH (fluid) + H30+K_1 (Ms) = 2H20 (11) due to the projection through Ti02). When projected into this diagram, the compositions of the heterogenous matrix biotite favoring the operation of the hydronium substitution in overlap the aluminous biotite end-member. these micas. More precise compositional data are needed to evaluate the relative importance of the pyrophyllite and hydronium substitutions. tions. These results would indicate rapid growth for the to po tactic and chessboard textures of the And + Kfs Muscovite breakdown + Bt pseudomorphs developed in the pegmatitic mus- Rubie and Brearley (1987) and Brearley and Rubie covite grains (Fig. 10), although the strongly overstepped (1990, e.g., their Fig. 5) observed topotactic and skeletal conditions for the Equilibrium 2 investigated by Rubie textures in muscovite that suffered disequilibrium break- and Brearley (1987) and Brearley and Rubie (1990) make down by metastable melting in laboratory experiments both cases not strictly comparable. However, Schramke (see also Brearley, 1986, for a natural case of metastable et al. (1987) obtained the stable assemblage And + Kfs breakdown). These authors obtained complete break- under moderate overstepping (0-100 °C, at 0.5-5 kbar) down of muscovite within 2-20 weeks for H20-saturated and H20-saturated conditions and concluded that Reac- and H20-undersaturated systems, under strongly over- tion 2 should quickly proceed to completion during pro- stepped experimental conditions (50-200 °C) at low P (1 grade metamorphism, even at conditions very close to kbar), and observed that the stable Kfs + Sil + Bt assem- the equilibrium boundary, if the rate-controlling step (a blage developed only under H20-undersaturated condi- surface-controlled rate involving andalusite surfaces; see ---~-

174 GARCIA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES also Kerrick et aI., 1991) remains unchanged through the ence of B or F (note the coexistence of tourmaline) or the reaction progress. Also, Ridley and Thompson (1986) es- infiltration from the matrix of a fluid with a lowered aK+/ timated a wide range of temperature overstepping re- aH+ratio (Wintsch, 1975, his Fig. 2) in the pegmatitic quired for nucleation in the breakdown of muscovite (Ms structure. + Qtz = Sil + Kfs + H20) as a function of interfacial energy and based on selected seeded and un seeded ex- CONCLUDING REMARKS perimental studies at 2 kbar, but they suggested the gen- The instability of the leucophyllite component in ma- eralized values of ca. 10 °C and 1 kbar for dehydration trix muscovite upon decompression is in accordance with reactions. These experimental and theoretical evidences other investigations on natural and experimental systems for low overstepping required for nucleation and high re- under sub solidus conditions. However, the high Si values action rates conflict with the fact that Reaction 2 did not found in the primary muscovite of the TGC stand as progress to completion in most of the observed pegma- exceptional when compared with typical compositions of titic muscovite crystals. But perhaps even more surpris- high-grade muscovite (e.g., Si = 6.2-6.3; Miller et aI., ing is the fact that primary and recrystallized matrix mus- 1981). Further, both the estimated P-T conditions of ear- covite crystals do not bear any textural evidence for ly equilibration (600-650 °C, > 10 kbar) and the presence breakdown through Reaction 2, in spite of the presence of concordant and crosscutting aplites strongly suggest of incompatible fine-grained quartz intergrown within that primary muscovite within the banded gneisses had them. been in equilibrium with a melt during this stage. Ex- The interpretation of the sequential growth of fibrous amples of anatectic granites either emplaced or generated sillimanite and andalusite in terms of a change from the at high pressure and bearing high-Si muscovite are not sillimanite to the andalusite stability fields with the re- common. However, the intrusive anatectic leucogranites action progress is further complicated by the metastable emplaced in kyanite-bearing peraluminous granitic nature of tibrolitic sillimanite, which might have grown gneisses in the Higher Himalayas (Searle and Fryer, 1986) outside the sillimanite stability field (cf. Kerrick, 1990). also bear muscovite with high Si contents (6.48 atoms However, the decompression path followed by these rocks pfu), which is consistent with our observations and in- (evolving from the kyanite, to the sillimanite, to the an- terpretation of the TGC as an instance of high-pressure dalusite stability fields, see Fig. 11) is consistent with ini- melting within a markedly thickened crust. tial growth of fibrous sillimanite close to Equilibrium 2. The data presented in this paper indicate that the com- The progress of the reaction, involving andalusite, must positional behavior of high- T muscovite is P-sensitive, also have taken place close to the equilibrium boundary, resembling that of lower grade environments, although otherwise muscovite would have reacted completely very high P conditions and high decompression rates within the andalusite stability field, according to the ex- would possibly be required to detect this effect in natural perimental evidence cited above. This and the fact that instances because of the increased probability that com- the amount of overstepping of Equilibrium 2 was not plete reequilibration takes place at low pressure during adequate for the activation of nucleation in the primary high-T emplacement or exhumation of granites and an- and recrystallized matrix crystals indicate that cooling ac- atectites. In the P- T-t path, section radiometric evidence companied decompression at this stage of the P-T evo- was summarized, indicating that high decompression rates lution and that the P-Tpath evolved close to Equilibrium characterized the low-P portion « 2-3 kbar) of the P-T 2. Additionally, kinetic factors affecting the reaction path depicted in Figure 11. Although more difficult to pathway (e.g., a change from surface-controlled to diffu- evaluate quantitatively, the reaction textures, and specif- sion-controlled growth?) could help explain the lack of ically the evidence for continuous nucleation and the complete breakdown. However, to explain the contrast- preservation of relict Si-rich compositions in primary ing behavior of matrix and pegmatitic muscovite crystals, muscovite from the TGC, may be taken as a strong in- modest displacements in the P- T space of Equilibrium 2 dication that high decompression rates were also preva- due to compositional differences of the phases, including lent during the earlier stage. This makes conceivable that the fluid, seem to be required. The virtually identical the age of the pressure peak in the Alpujarride series would compositions of recrystallized and pegmatitic grains in not have been far in time from the low-P cooling ages the analyzed samples (Fig. 4) and the fact that muscovite clustering at 21 :!: 2 Ma. In this regard it is worth men- breakdown textures have been observed only in aplitic tioning that a 4°Ar/39Ar age of 25.4 :!: 0.4 Ma was re- and pegmatitic bodies within the TGC suggest that vari- ported by Monie et al. (1991b), after a low-T phengite ations in the local fluid composition could be the major coexisting with carpholite, aragonite, and chloritoid in a factor affecting the onset of breakdown. Important com- sample from the Alpujarride Trevenque unit (located 50 positional variables of the local fluids in equilibrium with km to the northeast of the TGC), which was meant to these rocks include the activity of ionic species (i.e., K +, mark the "end of the underthrusting processes in the AI- Na+, H+; e.g., Wintsch, 1975) and ofB and F (Manning pujarride nappes and the beginning of exhumation." If and Pichavant, 1983; Pichavant, 1987). The onset of the that were the case, an average uplift rate of about 10 km/ breakdown of the pegmatitic muscovite should be pro- m.y. is easily derived for the whole decompression tra- moted at higher P by a lowered aH20caused by the pres- jectory of Figure 11, which illustrates what kind of ex- GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-T GNEISSES 175 treme geological conditions accompanied the production Fletcher, C,J.N., and Greenwood, H.J. (1979) Metamorphism and struc- of the muscovite textures reported in this study. ture of the Penfold Creek Area, near Quesnel Lake, British Columbia. Journal of Petrology, 20, 743-794. Forbes, W.C., and Flower, M.F.J. (1974) Phase relations of titan-phlog- ACKNOWLEDGMENTS opite, K2M~ TiAI2Si702o(OH)4: A refractory phase in the upper mantle? We wish to thank C.V. Guidotti, D.R. Veblen, and H,J. Massonne for Earth and Planetary Science Letters, 22, 60-66. their suggestions and comments on an earlier version of the manuscript, Fortier, S.M., and Giletti, B.J. (1991) Volume self-diffusion of oxygen in and J.T. Cheney, G. Guthrie, and Associate Editor J.B. Brady for their biotite, muscovite, and phlogopite micas. Geochimica et Cosmochimi- constructive reviews, which contributed substantially to the article's im- ca Acta, 55, 1319-1330. provement. Thanks are also given to D.S. Silverberg for his comments Franz, G., Thomas, S., and Smith, D.C. (1986) High-pressure phengite and help with the English version and M.A. Hidalgo Laguna for his careful decomposition in the Weissenstein eclogite, Miinchberger Gneiss Mas- attention to the microprobe analyses. The results presented in this paper sif, Germany. Contributions to Mineralogy and Petrology, 92, 71-85. form part of a Ph.D. thesis by A.G.C. This study has received financial Ganguly, J., and Saxena, S.K. (1984) Mixing properties of alumino silicate support from the Spanish CAICYT (project PB89-0017, A.G.C. and garnets: Constraints from natural and experimental data, and applica- R.L.T.R., and project PB87-022801, A.S.N.) and from the Junta de An- tions to geothermo-barometry. American Mineralogist, 69, 88-97. dalucia (research groups 4065, A.S.N., and 4072, A.G.C. and R.L.T.R.). Ghent, E.D., and Stout, M.Z. (1981) Geobarometry and geothermometry of plagioclase-biotite-garnet-muscovite assemblages. Contributions to Mineralogy and Petrology, 76, 92-97. REFERENCES CITED Gonzalez-Donoso, J.M., Linares, D., Molina, E., Serrano, F., and Vera, Abrecht, J., and Hewitt, D.A. (1988) Experimental evidence on the sub- J .A. (1982) Sobre la edad de la Formacion de la Vinuela (Cordilleras stitution ofTi in biotite. American Mineralogist, 73, 1275-1284. Beticas, Provincia de Malaga). Boletin de la Real Sociedad Espanola Berman, R.G. (1990) Mixing properties of Ca-Mg-Fe-Mn garnets. Amer- de Historia Natural, 80, 255-275. ican Mineralogist, 75, 328-344. Green, T.H. (1977) Garnet in silicic liquids and its possible use as a P- T Brady, J.B., and Stout, J.H. (1980) Normalization of thermodynamic indicator. Contributions to Mineralogy and Petrology, 65, 59-67. properties and some implications for graphical and analytical problems Greenwood, H.J. (1975) Thermodynamically valid projections of exten- in petrology. American Journal of Science, 280, 173-189. sive phase relationships. American Mineralogist, 60, 1-8. Brearley, A.J. (1986) An electron optical study of muscovite breakdown Guidotti, C.V. (1973) Compositional variation in muscovite as a function in pelitic xenoliths during pyrometamorphism. Mineralogical Maga- of metamorphic grade and assemblage in metapelites from N. W. Maine. zine, 50, 385-397. Contributions to Mineralogy and Petrology, 42, 33-42. Brearley, A.J., and Rubie, D.C. (1990) Effects of H20 on the disequilib- - (1978a) Compositional variation of muscovite in medium- to high- rium breakdown of muscovite + quartz. Journal of Petrology, 31, 925- grade metapelites of northwestern Maine. American Mineralogist, 63, 956. 878-884. Chatterjee, N.D., and Froese, E. (1975) A thermodynamic study of the - (1978b) Muscovite and K-feldspar from two-mica adamellite in pseudobinary join muscovite-paragonite in the system KAISi30s- northwestern Maine: Composition and petrogenetic implications. NaAISi30s-AI203-Si02-H20. American Mineralogist, 60, 985-993. American Mineralogist, 63, 750-753. Chatterjee, N.D., and Johannes, W. (1974) Thermal stability and standard -(1984) Micas in metamorphic rocks. In Mineralogical Society of thermodynamic properties of synthetic 2M] -muscovite, America Reviews in Mineralogy, 13, 357-468. KAI2[AISi30IO(OH)2]. Contributions to Mineralogy and Petrology, 48, Guidotti, C.V., and Sassi, F.P. (1976) Muscovite as a petrogenetic indi- 89-114. cator mineral in pelitic schists. Neues Jahrbuch flir Mineralogie Ab- Cipriani, C., Sassi, F.P., and Scolari, A. (1971) Metamorphic white micas: handlungen, 127, 97-142. Definition of paragenetic fields. Schweizerische mineralogische und pe- Guthrie, G.D., Jr., and Veblen, D.R. (1989) High-resolution transmission trographische Mitteilungen, 51, 259-302. electron microscopy of mixed-layer illite/smectite: Computer simula- Cuevas, J., Navarro-Vila, F., and Tubia, J.M. (1989) Interpretation des tions. Clays and Clay Minerals, 37, 1-11. cisaillements ductiles vers Ie NE dans les gneiss de Torrox (Complexe - (1990) Interpreting one-dimensional high-resolution transmission Alpujarride, Cordilleres Betiques). Geodinamica Acta, 3, 107-116. electron micrographs of sheet silicates by computer simulations. Amer- Dyar, M.D., Colucci, M.T., and Guidotti, C.V. (1991) Forgotton major ican Mineralogist, 75, 276-288. elements: Hydrogen and oxygen variation in biotite from metapelites. Heinrich, C.A. (1982) Kyanite-eclogite to amphibolite facies evolution of Geology, 19, 1029-1032. hydrous mafic and pelitic rocks. Adula nappe, Central Alps. Contri- Dymek, R.F. (1983) Titanium, aluminum and interlayer cation substi- butions to Mineralogy and Petrology, 81, 30- 38. tutions in biotite from high-grade gneisses, West Greenland. American Hewitt, D.A., and Abrecht, J. (1986) Limitations on the interpretation of Mineralogist, 68, 880-899. biotite substitutions from chemical analyses of natural samples. Amer- Eggleton, R.A., and Buseck, P.R. (1980) High resolution electron micros- ican Mineralogist, 71, 1126-1128. copy of feldspar weathering. Clays and Clay Minerals, 28, 173-178. Hodges, K.V., and Crowley, P.D. (1985) Error estimation and empirical Elorza, J.J., and Garcia-Duenas, V. (1981) Hoja y memoria explicativa geobarometry for pelitic systems. American Mineralogist, 70, 702-709. de Velez-Malaga. Mapa Geologico de Espana, escala 1:50000. Instituto Hodges, K.V., and Spear, F.S. (1982) Geothermometry, geobarometry, Geologico y Minero de Espana, Madrid. and the Al2SiOs triple point at Mt. Moosilauke, New Hampshire. Ernst, W.G. (1963) Significance ofphengitic micas from low grade schists. American Mineralogist, 67, 1118-1134. American Mineralogist, 48, 1357-1373. Hoisch, T.D. (1990) Empirical calibration of six geobarometers for the Evans, B.W., and Guidotti, C.V. (1966) The sillimanite-potash feldspar mineral assemblage quartz + muscovite + biotite + plagioclase + isograd in western Maine, USA. Contributions to Mineralogy and Pe- garnet. Contributions to Mineralogy and Petrology, 104, 225-234. trology, 12, 25-62. Holdaway, M.J. (1971) Stability ofandalusite and the aluminum-silicate Evans, B.W., and Patrick, B.E. (1987) Phengite-3Tin high-pressure meta- phase diagram. American Journal of Science, 271, 97-131. morphosed granitic orthogneisses, Seward Peninsula, Alaska. Canadian Holdaway, M.J., Dutrow, B.L., and Hinton, R.W. (1988) Devonian and Mineralogist, 25, 141-158. Carboniferous metamorphism in west-central Maine: The muscovite- Ferrow, E.A., London, D., Goodman, K.S., and Veblen, D.R. (1990) Sheet almandine geobarometer and the staurolite problem revisited. Ameri- silicates of the Lawler Peak granite, Arizona: Chemistry, structural can Mineralogist, 73, 20-47. variations, and exsoh.J.tion. Contributions to Mineralogy and Petrology, Iijima, S., and Zhu, J. (1982) Electron microscopy of a muscovite-bioite 105, 491-501. interface. American Mineralogist, 67, 1195-1205. Ferry, J.M., and Spear, F.S. (1978) Experimental calibration of the par- Indares, A., and Martignole, J. (1985) Biotite-garnet geothermometry in titioning of Fe and Mg between biotite and garnet. Contributions to granulite facies: The influence of Ti and Al in biotite. American Min- Mineralogy and Petrology, 66, 113-117. eralogist, 70, 272-278. ------

176 GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-TGNEISSES

Kerrick, D.M. (1972) Experimental determination of muscovite + quartz Pouchou, J.L., and Pichoir, F. (1985) "PAP" c/>(pz)procedure for im- stability with PH20< Ptotal'American Journal of Science, 272, 946-958. proved quantitative microanalysis. In J.T. Armstrong, Ed., Microbeam - (1990) The Al2SiOs polymorphs. Mineralogical Society of America Analysis, p. 104. San Francisco Press, San Francisco. Reviews in Mineralogy, 22, 406 p. Price, R.C. (1983) Geochemistry of a peraluminous granitoid suite from Kerrick, D.M., Lasaga, A.C., and Raeburn, S.P. (1991) Kinetics of het- north-eastern Victoria, south-eastern Australia. Geochimica et Cos- erogeneous reactions. In Mineralogical Society of America Reviews in mochimica Acta, 47,31-42. Mineralogy, 26, 583-671. Putnis, A., and McConnell, J.D.C. (1980) Principles of mineral behavior, Kistler, R.W., Ghent, E.D., and O'Neil, J.R. (1981) Petrogenesis of garnet 257 p. Blackwell, Oxford. two-mica granites in the Ruby Mountains, Nevada. Journal of Geo- Ridley, J. (1985) The effect of reaction enthalpy on the progress of a physical Research, 86, 10591-10606. metamorphic reaction. Advances in Physical Geochemistry, 4, 80-97. Konings, RJ.M., Boland, J.N., Vriend, S.P., and Jansen, J.B.H. (1988) Ridley, J., and Thompson, A.B. (1986) The role of mineral kinetics in Chemistry of biotites and muscovites in the Abas granite, northern the development of metamorphic microtextures. Advances in Physical Portugal. American Mineralogist, 73, 754-765. Geochemistry, 5, 154-193. Kretz, R. (1983) Symbols for rock-forming minerals. American Miner- Robinson, P., Jaffe, H.W., Ross, M., and Klein, C., Jr. (1971) Orientation alogist, 68, 277-279. of exsolution lamellae in c1inopyroxenes and clinoamphiboles: Consid- Labotka, T.C. (1983) Analysis of compositional variations of biotite in eration of optimal phase boundaries. American Mineralogist, 56, 909- pelitic hornfelses from northeastern Minnesota. American Mineralo- 939. gist, 68, 900-914. Rosenberg, P.H. (1987) Synthetic muscovite solid solutions in the system Lasaga, A.C. (1981) The atomistic basis of kinetics: Defects in minerals. K20-AI203-Si02-H20. American Mineralogist, 72, 716-723. In Mineralogical Society of America Reviews in Mineralogy, 8, 261- Rubie, D.C., and Brearley, A.J. (1987) Metastable melting during the 319. breakdown of muscovite + quartz at 1 kbar. Bulletin de Mineralogie, Lee, D.E., Kistler, R.W., Friedman, I., and Van Loenen, R.E. (1981) Two- 110, 533-549. mica granites of northeastern Nevada. Journal of Geophysical Re- Saliot, P., and Velde, B. (1982) Phengite compositions and post-nappe search, 86, 10607-10616. high-pressure metamorphism in the Pennine zone of the French Alps. Loomis, T.P. (1976) Irreversible reactions in high-grade metapelitic rocks. Earth and Planetary Science Letters, 57, 133-138. Journal of Petrology, 17, 559-588. Sanz de Galdeano, C. (1989) Estructura de las Sierras de Tejeda y de - (1979) A natural example of metastable reactions involving garnet Competa (Conjunto Alpujarride, Cordilleras Beticas). Revista de la So- and sillimanite. Journal of Petrology, 20, 271-292. ciedad Geologica de Espana, 2, 77-84. Manning, D.A.C., and Pichavant, M. (1983) The role of fluorine and Schramke, J.A., Kerrick, D.M., and Lasaga, A.C. (1987) The reaction boron in the generation of granitic melts. In M.P. Atherton and C.D. muscovite + quartz = andalusite + K-feldspar + water. Part 1. Growth Gribble, Eds., Migmatites, melting and metamorphism, p. 94-109. Shiva kinetics and mechanism. American Journal of Science, 287, 517-559. Publishing, Nantwich, England. Searle, M.P., and Fryer, B.J. (1986) Gamet, tourmaline and muscovite- Massonne, H.J., and Schreyer, W. (1986) High-pressure syntheses and bearing leucogranites, gneisses and migmatites of the Higher Himalayas X-ray properties of white micas in the system K20-MgO-AI203-Si02- from Zanskar, Kulu, Lahoul and Kashmir. The Geological Society Spe- H20. Neues Jahrbuch flir Mineralogie Abhandlungen, 153, 177-215. cialPublication, 19, 185-201. -(1987) Phengite geobarometry based on the limiting assemblage Sevigny, J.H., Parrish, R.R., and Ghent, E.D. (1989) Petrogenesis of per- with K-feldspar, phlogopite, and quartz. Contributions to Mineralogy aluminous granites, Monashee Mountains, southern Canadian Cordil- and Petrology, 96, 212-224. lera. Journal of Petrology, 30, 557-581. Miller, C.F., Stoddard, E.F., Bradfish, LJ., and Dollase, W.A. (1981) Speer, J.A. (1982) Metamorphism of pelitic rocks of the Snyder Group in Composition of plutonic muscovite: Genetic implications. Canadian the contact aureole of the Kigalpait layered intrusion, Labrador: Effects Mineralogist, 19, 25-34. of buffering partial pressures of water. Canadian Journal of Earth Sci- Miyashiro, A., and Shido, F. (1985) Tschermak substitution in low- and ences, 19, 1888-1909. middle-grade pelitic schists. Journal of Petology, 26, 449-487. -(1984) Micas in igneous rocks. In Mineralogical Society of Amer- Monie, P., Torres-Roldan, R.L., Garcia-Casco, A., and Goffe, B. (1991a) ica Reviews in Mineralogy, 13, 229-356. High rates of cooling in the western Alpujarrides, Betic Cordilleras, Thompson, A.B. (1976) Mineral reactions in pelitic rocks II. Calculation southern Spain: A 4°Arf39Ar study. Terra Nova Abstract, 3 (suppl. 6), 8. of some P-T-X (Fe-Mg) phase relations. American Journal of Science, Monie, P., Galindo-Zaldivar, J., Gonzalez-Lodeiro, F., Goffe, B., and 276, 425-454. Javaloy, A. (1991 b) 4°Ar/39Ar geochronology of Alpine tectonism in the - (1982) Dehydration melting of pelitic rocks and the generation of Betic Cordilleras (southern Spain). Journal of the Geological Society of H20-undersaturated granitic liquids. American Journal of Science, 282, London, 148, 288-297. 1567-1595. Monier, G., and Robert, J.L. (1986a) Muscovite solid solutions in the Thompson, A.B., and Ridley, J.R. (1987) Pressure-temperature-time (P- system K20-MgO-FeO-AI203-Si02-H20: An experimental study at 2 T-t) histories of orogenic belts. Philosophical Transactions of the Royal kbar PH20 and comparison with natural Li-free white micas. Miner- Society of London, A 321,27-45. alogical Magazine, 50, 257-266. Thompson, J.B., Jr (1979) The Tschermak substitution and reaction in - (1986b) Titanium in muscovites from two mica granites: Substi- pelitic schists. In V.A. Zharikov, V.I. Fonorev, and J.P. Korikovskii, tutional mechanisms and partition with coexisting biotites. Neues Jahr- Eds., Problems in physicochemical petrology, p. 146-159. Academy of buch flir Mineralogie Abhandlungen, 153, 147-161. Sciences, Moscow (in Russian). Monier, G., Mergoil-Daniel, J., and Labernardiere, H. (1984) Generations - (1982a) Compositional space: An algebraic and geometric ap- succesives de muscovites et feldspaths potassiques dans les leucogra- proach. In Mineralogical Society of America Reviews in Mineralogy, nites du massif de Millevaches (Massif Central fran~ais). Bulletin de 10,1-31. Mineralogie, 107, 55-68. -(1982b) Reaction space: An algebraic and geometric approach. In Newton, R.C., and Haselton, H.T. (1981) Thermodynamics of the garnet- Mineralogical Society of America Reviews in Mineralogy, 10, 33-52. plagioclase-AI2SiOs-quartz geobarometer. Advances in Physical Geo- Torres-Roldan, R.L. (1974) El metamorfismo progresivo y la evolucion chemistry, 1, 129-145. de la serie de facies en las metapelitas alpujarrides al SE de Sierra Perchuk, L.L., and Larent'eva, LV. (1983) Experimental investigation of Almijara (sector central de las Cordilleras Beticas. S de Espana). Cua- exchange equilibria in the system cordierite-garnet-biotite. Advances demos de Geologia, 5,21-77. in Physical Geochemistry, 3, 199-239. - (1979) The tectonic subdivision of the Betic Zone (Betic Cordil- Pichavant, M. (1987) Effects of B and H20 on liquidus phase relations in lera, S. Spain): Its significance and one possible geotectonic scenario the haplogranite system at 1 kbar. American Mineralogist, 72, 1056- for the westernmost Alpine Belt. American Journal of Science, 279, 1070. 19- 51. GARCiA-CASCO ET AL.: MUSCOVITE IN HIGH P-TGNEISSES 177

-(1981) Plurifacial metamorphic evolution of the Sierra Bermeja Weiss, Z., Rieder, M., Chmielova, M., and Krajicek, J. (1985) Geometry peridotite aureole (southern Spain). Estudios Geol6gicos, 37, 115-133. of the octahedral coordination in micas: A review of refined structures. Tracy, R. (1978) High grade metamorphic reactions and partial melting American Mineralogist, 70, 747-757. in pelitic schists, west-central Massachusetts. American Journal ofSci- Wintsch, R.P. (1975) Solid-fluid equilibria in the system KAISi30g- ence, 278, 150-178. NaAISi30g-AI2SiOs-Si02-H20-HCl. Journal of Petrology, 16, 57-79. Veblen, D.R. (1983) Microstructures and mixed layering in intergrown Wise, W.S., and Eugster, H.P. (1964) Celadonite: Synthesis, thermal sta- wonesite, chlorite, talc, biotite, and kaolinite. American Mineralogist, bility and occurrence. American Mineralogist, 49, 1031-1083. 68, 566-580. Zeck, H.P., Albat, F., Hansen, B.T., Torres-Roldan, R.L., Garcia-Casco, Velde, B. (1965) Phengite micas: Synthesis, stability, and natural occur- A., and Martin-Algarra, A. (1989) A 21 :t 2 Ma age for the termination rence. American Journal of Science, 263, 886-913. of the ductile Alpine deformation in the internal zone of the Betic - (1967) Si+4content of natural phengites. Contributions to Miner- Cordilleras, south Spain. Tectonophysics, 169,215-220. alogy and Petrology, 14, 250-258. Zeck, H.P., Monie, P., Villa, I., and Hansen, B.T. (1992) Very high rates - (1969) The compositional join muscovite-pyrophyllite at moder- of cooling and uplift in the Alpine belt of the Betic Cordilleras, southern ate pressures and temperatures. Bulletin de la Societe Franyaise de Mi- Spain. Geology, 20, 79-82. neralogie et de Cristallographie, 92, 360-368. Zussman, J. (1979) The crystal chemistry of the micas. Bulletin de Mi- Wang, G.F., and Banno, S. (1987) Non-stoichiometry of interlayer ca- neralogie, 102, 5-13. tyions in micas from low- to middle-grade metamorphic rocks in the Ryoke and the Sanbagawa belts, Japan. Contributions to Mineralogy MANUSCRIPT RECEIVED SEPTEMBER 9, 1991 and Petrology, 97, 313-319. MANUSCRIPT ACCEPTED AUGUST 26, 1992